Fault diagnosis method and device, storage medium and electronic equipment
By combining network signal sequences, fault codes, and topology information, and using LSTM, BERT, and GCN models for feature extraction and path generation, the problem of high-precision real-time diagnosis of vehicle network faults in distributed gateway scenarios is solved, enabling accurate identification and prediction of vehicle network faults.
Patent Information
- Application Number
- CN202511166144.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-08-20
- Publication Date
- 2025-11-11
AI Technical Summary
In distributed gateway scenarios, it is difficult to achieve high-precision real-time diagnosis of vehicle network faults, especially when the network load changes dynamically, existing technologies are unable to effectively identify and predict faults.
By acquiring network signal sequences and fault codes uploaded by the central brain, and combining signal temporal features, fault semantic features, and network topology information, LSTM, BERT, and GCN models are used for feature extraction and path generation to achieve high-precision network fault diagnosis.
High-precision real-time fault diagnosis was achieved in complex network environments with multiple gateways. It can identify known faults and predict potential faults, thereby improving the accuracy and efficiency of fault diagnosis.
Smart Images
Figure CN120935218A_ABST
Abstract
Description
Technical Field
[0001] This application belongs to the field of vehicle inspection technology, and in particular relates to a fault diagnosis method, device, storage medium and electronic equipment. Background Technology
[0002] As vehicle intelligence continues to develop, the vehicle's electronic and electrical architecture is also constantly evolving. In distributed gateway scenarios, a central brain corresponds to multiple regional gateways, forming a complex network environment with multiple gateways and subnets.
[0003] Diagnosing network faults in vehicles requires analyzing network fault chains using vehicle signals. However, vehicle signals and network topology are complex, making it difficult to achieve high-precision real-time fault diagnosis under dynamically changing network loads, ultimately affecting vehicle operation. Summary of the Invention
[0004] The embodiments of this application provide a fault diagnosis method, device, storage medium, and electronic device, which can combine network signal sequences, network fault codes, and network topology information to comprehensively analyze the network fault status of a target vehicle, thereby achieving high-precision real-time fault diagnosis in complex network environments with multiple gateways.
[0005] Other features and advantages of this application will become apparent from the following detailed description, or may be learned in part from practice of this application.
[0006] According to a first aspect of the embodiments of this application, a fault diagnosis method is provided, applied to the cloud corresponding to the gateway system of a target vehicle. The gateway system includes a central brain and at least two regional gateways. The fault diagnosis method includes:
[0007] Obtain network signal sequences and network fault codes uploaded by the central brain;
[0008] Feature extraction is performed on the network signal sequence to obtain signal temporal features; these features are used to characterize the signal variation trend of the network signal sequence.
[0009] Feature extraction is performed on network fault codes to obtain fault semantic features; among which, fault semantic features are used to characterize the degree of correlation between fault objects corresponding to network fault codes;
[0010] Generate a signal propagation path based on the network topology information of the gateway system;
[0011] Based on signal timing characteristics, fault semantic characteristics, and signal propagation paths, network fault diagnosis is performed on the target vehicle to obtain diagnostic results.
[0012] In some embodiments of this application, based on the aforementioned scheme, feature extraction is performed on the network signal sequence to obtain signal time-series features, including:
[0013] Divide the network signal sequence into at least two signal subsequences of a preset time length;
[0014] At least two signal subsequences are input into the signal analysis model to extract features from the at least two signal subsequences, thereby obtaining the signal time-series features output by the signal analysis model; wherein, the signal analysis model is constructed based on LSTM.
[0015] In some embodiments of this application, based on the aforementioned scheme, feature extraction is performed on network fault codes to obtain fault semantic features, including:
[0016] Obtain the descriptive text data corresponding to the network fault codes; the descriptive text data is used to define the fault objects corresponding to different network fault codes.
[0017] The network fault codes and descriptive text data are input into the fault analysis model to extract features from the network fault codes based on the descriptive text data, thereby obtaining the fault semantic features output by the fault analysis model; the fault analysis model is built based on BERT.
[0018] In some embodiments of this application, before generating a signal propagation path based on the network topology information of the gateway system according to the foregoing scheme, the method further includes:
[0019] Using the central brain and regional gateways as nodes, and the first and second communication links as edges, a knowledge graph of the gateway system is constructed; wherein, the first communication link is the communication link between the central brain and the regional gateways, and the second communication link is the communication link between the regional gateways.
[0020] From the knowledge graph of the gateway system, the network topology information corresponding to the different vehicle functions involved in the gateway system is extracted.
[0021] In some embodiments of this application, based on the foregoing scheme, a signal propagation path is generated according to the network topology information of the gateway system, including:
[0022] For each vehicle function involved in the gateway system, the network topology information is input into the graph structure recognition model to obtain the signal propagation path under the vehicle function output by the graph structure recognition model. The graph structure recognition model is built based on GCN, and the signal propagation path includes the regional gateway associated with the vehicle function.
[0023] In some embodiments of this application, based on the aforementioned scheme, network fault diagnosis is performed on the target vehicle according to signal timing characteristics, fault semantic characteristics, and signal propagation path to obtain diagnostic results, including:
[0024] The signal propagation path is converted into a text sequence, and features are extracted from the text sequence to obtain the signal path features;
[0025] The signal timing features, fault semantic features, and signal path features are normalized, and then the normalized signal timing features, fault semantic features, and signal path features are weighted and fused to obtain the fault probability information corresponding to the central brain and each regional gateway.
[0026] In some embodiments of this application, based on the foregoing scheme, the method further includes:
[0027] In response to a diagnostic result request for the target vehicle sent by the target device, determine the corresponding result feedback strategy based on the device type of the target device;
[0028] Fault warning information is generated based on the result feedback strategy and sent to the target device.
[0029] According to a second aspect of the embodiments of this application, a fault diagnosis device is provided, configured in the cloud corresponding to the gateway system of a target vehicle. The gateway system includes a central brain and at least two regional gateways. The fault diagnosis device includes:
[0030] The information acquisition module is used to acquire network signal sequences and network fault codes uploaded by the central brain.
[0031] The first feature extraction module is used to extract features from the network signal sequence to obtain signal temporal features; wherein, the signal temporal features are used to characterize the signal change trend of the network signal sequence.
[0032] The second feature extraction module is used to extract features from network fault codes to obtain fault semantic features; wherein, fault semantic features are used to characterize the degree of correlation between fault objects corresponding to network fault codes.
[0033] The path generation module is used to generate signal propagation paths based on the network topology information of the gateway system.
[0034] The fault diagnosis module is used to perform network fault diagnosis on the target vehicle based on signal timing characteristics, fault semantic characteristics, and signal propagation path, and obtain the diagnosis results.
[0035] According to a third aspect of the embodiments of this application, a computer-readable storage medium is provided, which stores computer program instructions that, when loaded and executed by a processor, implement the steps of the method as described in any of the first aspects above.
[0036] According to a fourth aspect of the embodiments of this application, an electronic device is provided, including a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, it implements the steps of the method as described in any of the first aspects above.
[0037] According to a fifth aspect of the embodiments of this application, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps of the method as described in any of the first aspects above.
[0038] In this application, the gateway system of the target vehicle includes a central brain and at least two regional gateways. The cloud-based gateway system acquires network signal sequences and network fault codes uploaded by the central brain. Feature extraction is performed on the network signal sequences to obtain signal temporal features characterizing the signal change trends. Feature extraction is also performed on the network fault codes to obtain fault semantic features characterizing the correlation between the fault objects corresponding to the fault codes. Based on the network topology information of the gateway system, a signal propagation path is generated. Then, based on the signal temporal features, fault semantic features, and signal propagation path, network fault diagnosis is performed on the target vehicle to obtain a diagnostic result. The technical solution provided in this application combines network signal sequences, network fault codes, and network topology information to comprehensively analyze the network fault situation of the target vehicle, thereby achieving high-precision real-time fault diagnosis in complex network environments with multiple gateways.
[0039] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description
[0040] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application. It is obvious that the drawings described below are merely some embodiments of this application, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort. In the drawings:
[0041] Figure 1 A schematic diagram of a scenario in which the fault diagnosis method of the embodiments of this application can be applied is shown;
[0042] Figure 2 A flowchart of a fault diagnosis method applied to the cloud in an embodiment of this application is shown;
[0043] Figure 3 A detailed flowchart of the extraction of signal timing features in an embodiment of this application is shown;
[0044] Figure 4 A detailed flowchart of extracting fault semantic features in an embodiment of this application is shown;
[0045] Figure 5 A detailed flowchart illustrating the generation of network topology information in an embodiment of this application is shown;
[0046] Figure 6 A detailed flowchart of network fault diagnosis in an embodiment of this application is shown;
[0047] Figure 7 A detailed flowchart of the fault warning implementation in an embodiment of this application is shown;
[0048] Figure 8 Another flowchart of the fault diagnosis method in the embodiments of this application is shown;
[0049] Figure 9 A block diagram of a fault diagnosis device configured in the cloud is shown in an embodiment of this application;
[0050] Figure 10 A schematic diagram of the structure of an electronic device in an embodiment of this application is shown. Detailed Implementation
[0051] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0052] Furthermore, the described features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. Numerous specific details are provided in the following description to give a thorough understanding of embodiments of this application. However, those skilled in the art will recognize that the technical solutions of this application can be practiced without one or more of the specific details, or other methods, components, apparatuses, steps, etc., can be employed. In other instances, well-known methods, apparatuses, implementations, or operations are not shown or described in detail to avoid obscuring various aspects of this application.
[0053] The block diagrams shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically independent entities. That is, these functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor devices and / or microcontroller devices.
[0054] The flowcharts shown in the accompanying drawings are merely illustrative and do not necessarily include all content and operations / steps, nor do they necessarily have to be performed in the described order. For example, some operations / steps can be broken down, while others can be combined or partially combined; therefore, the actual execution order may change depending on the specific circumstances.
[0055] To enable those skilled in the art to better understand this application, firstly, in conjunction with Figure 1 A brief description of the application scenarios involved in this application is provided.
[0056] See Figure 1 The diagram illustrates a scenario where the fault diagnosis method of the embodiments of this application can be applied.
[0057] The target vehicle's gateway system includes a central brain 101 and multiple regional gateways 102 and 103. The central brain 101 has certain data processing capabilities and can communicate with each regional gateway via the network, coordinating the functions of the corresponding regional controllers. The central brain 101 can also communicate with the relevant cloud 100, uploading signals collected by each regional gateway to the cloud 100 to enable network fault diagnosis of the target vehicle.
[0058] Specifically, each regional gateway, synchronized with the clock of the central brain 101, sends the collected signals to the central brain 101. The central brain 101 then forms a network signal sequence and uploads it to the cloud 100. The central brain 101 also generates network fault codes based on preliminary fault diagnosis and uploads these codes to the cloud 100 as well. The central brain 101 can directly send the network signal sequence and network fault codes to the cloud 100, or it can first upload them to the relevant database in the cloud 100 for use by the cloud 100 when performing network fault diagnosis on the target vehicle.
[0059] Correspondingly, Cloud 100 can obtain network signal sequences and network fault codes uploaded by Central Brain 101, extract features from the network signal sequences to obtain signal temporal features that characterize the signal change trend of the network signal sequences, extract features from the network fault codes to obtain fault semantic features that characterize the degree of correlation between the fault objects corresponding to the network fault codes, and generate signal propagation paths based on the network topology information of the gateway system. Then, based on the signal temporal features, fault semantic features, and signal propagation paths, network fault diagnosis is performed on the target vehicle to obtain the diagnosis results.
[0060] Cloud 100 can be implemented using a standalone server or a server cluster consisting of multiple servers. Cloud 100 can integrate a data storage system to maintain the database, which stores the network signal sequence and network fault codes of the target vehicle.
[0061] In one exemplary embodiment, refer to Figure 2 The flowchart of the fault diagnosis method in this application embodiment is shown. It is applied to the cloud corresponding to the gateway system of the target vehicle. The gateway system includes a central brain and at least two regional gateways, which are described in detail below:
[0062] Step 201: Obtain the network signal sequence and network fault code uploaded by the central brain.
[0063] Multiple regional gateways on the target vehicle collect vehicle signals in real time and send them to the central processing unit on the vehicle. The central processing unit then filters the vehicle signals from different regional gateways, generates a network signal sequence according to the order of collection time, and finally uploads the network signal sequence to the cloud. This network signal sequence can include various types of signal sequences, such as wake-up message sequences, network management message sequences, and sensor signal sequences.
[0064] The central processing unit on the target vehicle can also perform preliminary fault detection, generate corresponding network fault codes, and upload these codes to the cloud. Specifically, the network fault codes refer to DTCs (Diagnostic Trouble Codes), which can indicate information such as the vehicle's region, system, and category of the fault.
[0065] The cloud can directly receive network signal sequences and network fault codes sent by the central brain of the target vehicle, and then execute the fault diagnosis method provided in this embodiment. The cloud can also store the network signal sequences and network fault codes sent by the central brains of different vehicles in a relevant database. Then, when it is necessary to perform fault diagnosis on the target vehicle, the cloud can retrieve the network signal sequence and network fault code of the target vehicle from the database and execute the fault diagnosis method provided in this embodiment.
[0066] Step 202: Extract features from the network signal sequence to obtain the signal time sequence features.
[0067] Among them, signal time series features are used to characterize the signal change trend of network signal sequences.
[0068] For example, the network signal sequence is a wake-up signal sequence. Regional gateway A sends a wake-up signal to the central brain, which then sends wake-up signals to regional gateways B and C respectively. Further, regional gateway C sends a wake-up signal to regional gateway D. The wake-up signals between each regional gateway and the central brain constitute the wake-up signal sequence. Feature extraction from the wake-up signal sequence yields the corresponding signal temporal features. These features characterize the transmission order of the wake-up signal between each regional gateway and the central brain, reflecting the signal change trend of the wake-up signal sequence.
[0069] Optionally, features can be extracted from different types of network signal sequences using machine learning models or neural network models to obtain the corresponding time-series features of different types of signals.
[0070] Step 203: Extract features from the network fault codes to obtain fault semantic features.
[0071] Among them, fault semantic features are used to characterize the degree of correlation between the fault objects corresponding to network fault codes. Network fault codes refer to DTCs, and the fault objects are the vehicle area and system where the fault occurred, as indicated by the DTC.
[0072] The network fault codes generated based on preliminary fault detection can only provide general information such as the vehicle region, system, and category where the fault occurred. Further analysis of the network fault codes is needed to obtain the implicit semantic features of the fault.
[0073] The cloud can analyze network fault codes based on the standard format of network fault codes to extract fault semantic features, or it can use machine learning models or neural network models to extract features from network fault codes to obtain fault semantic features.
[0074] Optionally, all network fault codes of the target vehicle can be combined into a fault code set, and feature extraction can be performed on the fault code set to obtain fault semantic features.
[0075] Step 204: Generate a signal propagation path based on the network topology information of the gateway system.
[0076] Different vehicles may have different gateway systems. The gateway system of a target vehicle includes a central processing unit (CPU) and at least two regional gateways. The network topology information of the target vehicle's gateway system includes the network communication links between the CPU and each regional gateway. The cloud can obtain the network topology information of the target vehicle's gateway system, parse it, and generate corresponding signal propagation paths. These signal propagation paths can be used to identify the propagation paths of abnormal signals within the gateway system.
[0077] For example, the network topology information of a gateway system can be represented by a gateway topology diagram. The central brain and each regional gateway have a unique corresponding node in the gateway topology diagram. The ECUs (Electronic Control Units) associated with the central brain and each regional gateway can be used as attribute information for their respective nodes. The network communication links between the central brain and each regional gateway are the edges between the nodes in the gateway topology diagram. Based on the gateway topology diagram, possible signal propagation paths can be identified.
[0078] Optionally, the signal propagation paths can be classified according to vehicle functions to determine the regional gateways and ECUs involved in different vehicle functions, thereby determining the signal propagation paths corresponding to different vehicle functions.
[0079] Step 205: Based on the signal timing characteristics, fault semantic characteristics, and signal propagation path, perform network fault diagnosis on the target vehicle to obtain the diagnosis results.
[0080] Multimodal feature fusion is performed on the temporal features of the time-series layer, the fault semantic features of the semantic layer, and the signal propagation path of the spatial layer to comprehensively analyze the network fault problems of the target vehicle and obtain diagnostic results. The diagnostic results include the vehicle components that have been identified as faulty, as well as other vehicle components that are predicted to be affected and thus fail.
[0081] Optionally, a multimodal feature fusion model is employed to diagnose network faults in the target vehicle. This model comprises at least a three-layer neural network architecture: a sequential layer, a semantic layer, and a spatial layer. Relevant features are extracted from each layer, and the output layer then performs a weighted fusion of these three features based on an attention mechanism, ultimately outputting the diagnostic result. The diagnostic result can be represented in the form of a fault probability matrix, where each element corresponds to a vehicle component, and each element represents the fault probability of the corresponding vehicle component.
[0082] In this application, the gateway system of the target vehicle includes a central brain and at least two regional gateways. The cloud-based gateway system acquires network signal sequences and network fault codes uploaded by the central brain. Feature extraction is performed on the network signal sequences to obtain signal temporal features characterizing the signal change trends. Feature extraction is also performed on the network fault codes to obtain fault semantic features characterizing the correlation between the fault objects corresponding to the fault codes. Based on the network topology information of the gateway system, a signal propagation path is generated. Then, based on the signal temporal features, fault semantic features, and signal propagation path, network fault diagnosis is performed on the target vehicle to obtain a diagnostic result. The technical solution provided in this application combines network signal sequences, network fault codes, and network topology information to comprehensively analyze the network fault situation of the target vehicle, thereby achieving high-precision real-time fault diagnosis in complex network environments with multiple gateways.
[0083] Based on the above embodiments, in an exemplary embodiment, see [link to example]. Figure 3 The flowchart illustrating the extraction of signal timing features in an embodiment of this application is shown, specifically including:
[0084] Step 301: Divide the network signal sequence into at least two signal subsequences of a preset time length.
[0085] Step 302: Input at least two signal subsequences into the signal analysis model to extract features from at least two signal subsequences and obtain the signal time-series features output by the signal analysis model.
[0086] The network signal sequence is divided into at least two signal subsequences according to a sliding time window. The window length of the time window is the preset time length, such as 5 seconds. In this embodiment, the specific value of the preset time length is not limited.
[0087] Furthermore, all signal subsequences are input into a pre-trained signal analysis model to extract temporal features and obtain the corresponding signal temporal features. The signal analysis model is built based on LSTM (Long Short-Term Memory).
[0088] For example, an initial signal analysis model is constructed based on LSTM, and sample data of network signal sequences from different vehicles are used to train the initial signal analysis model until the training termination condition is met, thus obtaining the signal analysis model in this embodiment. Furthermore, when it is necessary to extract the signal temporal features of the target vehicle's network signal sequence, feature extraction is performed through the signal analysis model to obtain signal temporal features characterizing the signal change trend of the network signal sequence.
[0089] In this application, a signal analysis model is constructed based on LSTM, and then the signal temporal features of the network signal sequence are extracted through the signal analysis model, which helps to improve the accuracy of the feature extraction results.
[0090] Based on the above embodiments, in an exemplary embodiment, see [link to example]. Figure 4 The flowchart illustrating the extraction of fault semantic features in an embodiment of this application is shown, specifically including:
[0091] Step 401: Obtain the description text data corresponding to the network fault code.
[0092] Step 402: Input the network fault code and description text data into the fault analysis model to extract features from the network fault code based on the description text data, and obtain the fault semantic features output by the fault analysis model.
[0093] As is understandable, Network Fault Codes (DTCs) are defined based on automotive communication standards, and each character in a DTC has a unique meaning. The descriptive text data is used to define the fault object corresponding to each different DTC.
[0094] Network fault codes and descriptive text data are used as input data for the fault analysis model to extract semantic features and obtain the corresponding fault semantic features. The fault analysis model is built based on BERT (Bidirectional Encoder Representation from Transformers).
[0095] For example, an initial fault analysis model is constructed based on BERT, and sample data of network fault codes from different vehicles, as well as descriptive text data, are used to train the initial fault analysis model until the training termination condition is met, thus obtaining the fault analysis model in this embodiment. Furthermore, when it is necessary to extract the fault semantic features of the target vehicle's network fault codes, feature extraction is performed through the fault analysis model to obtain fault semantic features that characterize the degree of correlation between the fault objects corresponding to the network fault codes.
[0096] In this application, a fault analysis model is constructed based on BERT, and then the fault semantic features of network fault codes are extracted through the fault analysis model, which helps to improve the accuracy of feature extraction results.
[0097] Based on the above embodiments, in an exemplary embodiment, see [link to example]. Figure 5 The flowchart illustrating the detailed process of generating network topology information in an embodiment of this application is shown, specifically including:
[0098] Step 501: Construct a knowledge graph of the gateway system, using the central brain and regional gateways as nodes and the first and second communication links as edges.
[0099] Step 502: Extract the network topology information corresponding to the different vehicle functions involved in the gateway system from the knowledge graph of the gateway system.
[0100] In the gateway system of the target vehicle, the central brain and each regional gateway are entity nodes, and the ECUs associated with the central brain and each regional gateway are the entity attributes corresponding to the entity nodes. For example, a regional gateway A in the gateway system is an entity node in the knowledge graph, and ECU01 associated with regional gateway A is an entity attribute of that entity node.
[0101] The communication links between the central brain and each regional gateway are described as the first communication link, and the communication links between each regional gateway are described as the second communication link. The relationships between each entity node are determined by using the communication links as the extraction criteria, thereby obtaining the edges in the knowledge graph and obtaining the knowledge graph of the gateway system.
[0102] Furthermore, using vehicle function as the categorization standard, network topology information corresponding to different vehicle functions is extracted from the knowledge graph of the gateway system. For example, when regional gateway B sends a wake-up message to the central brain, the central brain can determine the ECU associated with regional gateway B, thereby identifying the relevant vehicle function. It then sends wake-up messages to other regional gateways associated with that vehicle function, such as the central brain sending a wake-up message to regional gateway C, activating the ECU associated with regional gateway C, and starting the vehicle function. The network topology information corresponding to this vehicle function is the network topology information contained in the communication link of regional gateway B - central brain - regional gateway C.
[0103] In this application, knowledge graphs are used to obtain network topology information of the gateway system. Compared with textual descriptions, image representations are more accurate and intuitive, and better reflect the actual state of the gateway system of the target vehicle. Furthermore, based on vehicle functions, network topology information corresponding to different vehicle functions can be obtained, which helps to improve the effect of generating signal propagation paths. Matching vehicle functions can be found based on signal types, thereby determining more accurate signal propagation paths.
[0104] Based on the above embodiments, in an exemplary embodiment, a signal propagation path is generated according to the network topology information of the gateway system, including: inputting the network topology information corresponding to each vehicle function involved in the gateway system into a graph structure recognition model to obtain the signal propagation path under the vehicle function output by the graph structure recognition model, wherein the signal propagation path includes the regional gateway associated with the vehicle function.
[0105] Specifically, for each vehicle function, the network topology information corresponding to that function is extracted from the knowledge graph of the gateway system. This includes the entity nodes involved in the vehicle function, the entity attributes corresponding to the entity nodes, and the relationships between the entity nodes. Since the network topology information is represented as an image, feature extraction can be performed using a graph structure recognition model to obtain the corresponding signal propagation path. The graph structure recognition model is built based on GCN (Graph Convolutional Network).
[0106] For example, an initial graph structure recognition model is constructed based on GCN, and sample data from the knowledge graphs of gateway systems of different vehicles are used to train the initial graph structure recognition model until the training termination condition is met, thus obtaining the graph structure recognition model in this embodiment. Furthermore, when it is necessary to generate a signal propagation path, feature extraction is performed through the graph structure recognition model to obtain the signal propagation path under the corresponding vehicle function.
[0107] In this application, a graph structure recognition model is constructed based on GCN, and then the signal propagation path is generated through the graph structure recognition model, which helps to improve the accuracy of the path generation results.
[0108] Based on the above embodiments, in an exemplary embodiment, see [link to example]. Figure 6 The flowchart illustrating the network fault diagnosis in this embodiment of the application is shown, specifically including:
[0109] Step 601: Convert the signal propagation path into a text sequence and extract features from the text sequence to obtain signal path features.
[0110] Step 602: Normalize the signal timing features, fault semantic features, and signal path features, and then perform weighted fusion processing on the normalized signal timing features, fault semantic features, and signal path features to obtain the fault probability information corresponding to the central brain and each regional gateway.
[0111] For example, network topology information can only represent the entity nodes involved in the vehicle function, the entity attributes corresponding to the entity nodes, and the relationships between entity nodes; it cannot represent the signal propagation path (including direction) between entity nodes. After obtaining the signal propagation path under the vehicle function through the graph structure recognition model, the signal propagation path represented by the image can be converted into a text sequence. The text sequence includes entity nodes, the entity attributes corresponding to the entity nodes, the relationships between entity nodes, and the propagation path (including direction) between entity nodes. Feature extraction is performed on the text sequence to obtain the corresponding signal path features.
[0112] It is understandable that signal timing features, fault semantic features, and signal path features are features under different dimensions. Before multimodal feature fusion, normalization processing is required. This embodiment does not limit the specific method of normalization processing.
[0113] Furthermore, based on experimentally determined weights for signal temporal features, fault semantic features, and signal path features, a weighted fusion calculation is performed on the aforementioned temporal layer features, semantic layer features, and spatial layer features using an attention mechanism to obtain fault probability information. This fault probability information can be a fault probability matrix, where each element corresponds to a vehicle component, and each element represents the fault probability of the corresponding vehicle component.
[0114] In this application, network fault diagnosis of the target vehicle is completed by multimodal feature fusion. It can not only identify known faults, but also infer new fault modes. By analyzing the fault probability of each node in the gateway system of the target vehicle, the effect of risk prediction is achieved.
[0115] Based on the above embodiments, in an exemplary embodiment, see [link to example]. Figure 7 The flowchart illustrating the fault warning process in this embodiment of the application is shown, specifically including:
[0116] Step 701: In response to the diagnostic result request for the target vehicle sent by the target device, determine the corresponding result feedback strategy based on the device type of the target device.
[0117] Step 702: Generate fault warning information based on the result feedback strategy and send the fault warning information to the target device.
[0118] The target equipment refers to intelligent devices related to the target vehicle, such as devices held by R&D personnel, after-sales personnel, and drivers. Different types of target equipment require different result feedback strategies. The result feedback strategy is used to indicate the information content, presentation format, and response mechanism for generating fault warning information.
[0119] For example, for equipment used by R&D personnel, fault warning information is generated, including but not limited to the original signal waveform, feature importance heatmap, and fault signal propagation path. This information can be presented as an interactive dashboard on a web platform, and the response mechanism can be the automatic creation of Jira work orders. For equipment used by after-sales personnel, fault warning information is generated, including but not limited to network fault code parsing, such as repair procedures (SOPs) and spare parts requirement lists. This information can be presented as mobile application work orders, and the response mechanism can be triggering the 4S store's ERP (Enterprise Resource Planning) system to prepare inventory. For equipment used by drivers, fault warning information is generated, including but not limited to central fault prompts, such as "Vehicle network abnormal, please restart," and navigation to the nearest service point. This information can be displayed on the vehicle screen or a mobile device, and the response mechanism can be in-vehicle voice broadcast and one-click appointment for repair.
[0120] In this application, corresponding fault warning methods are adopted for the equipment type of the target device, which can execute customized feedback processes for different objects, which is conducive to timely fault feedback of the target vehicle and improves the user experience.
[0121] To enable those skilled in the art to better understand this application as a whole, the application process of the solution in this application will be briefly described below with reference to a specific embodiment:
[0122] See Figure 8 This illustrates another flowchart of the fault diagnosis method in an embodiment of this application, which specifically includes:
[0123] Step 801: Obtain the network signal sequence and network fault code uploaded by the central brain.
[0124] Step 802: Divide the network signal sequence into at least two signal subsequences of a preset time length.
[0125] Step 803: Input at least two signal subsequences into the signal analysis model to extract features from the at least two signal subsequences and obtain the signal time-series features output by the signal analysis model.
[0126] Among them, the signal analysis model is built based on LSTM, and the signal temporal features are used to characterize the signal change trend of the network signal sequence;
[0127] Step 804: Obtain the description text data corresponding to the network fault code.
[0128] The descriptive text data is used to define the fault objects corresponding to different network fault codes.
[0129] Step 805: Input the network fault code and description text data into the fault analysis model to extract features from the network fault code based on the description text data, and obtain the fault semantic features output by the fault analysis model.
[0130] Among them, the fault analysis model is built based on BERT, and the fault semantic features are used to characterize the degree of correlation between the fault objects corresponding to the network fault codes.
[0131] Step 806: Construct a knowledge graph of the gateway system, using the central brain and regional gateways as nodes and the first and second communication links as edges.
[0132] The first communication link is the communication link between the central brain and each regional gateway, and the second communication link is the communication link between each regional gateway.
[0133] Step 807: Extract the network topology information corresponding to the different vehicle functions involved in the gateway system from the knowledge graph of the gateway system.
[0134] Step 808: For each vehicle function involved in the gateway system, input the network topology information into the graph structure recognition model to obtain the signal propagation path under the vehicle function output by the graph structure recognition model.
[0135] The graph structure recognition model is based on GCN, and the signal propagation path includes the regional gateway associated with vehicle functions.
[0136] Step 809: Convert the signal propagation path into a text sequence and extract features from the text sequence to obtain signal path features.
[0137] Step 810: Normalize the signal timing features, fault semantic features, and signal path features, and then perform weighted fusion processing on the normalized signal timing features, fault semantic features, and signal path features to obtain the fault probability information corresponding to the central brain and each regional gateway.
[0138] Step 811: In response to the diagnostic result request for the target vehicle sent by the target device, determine the corresponding result feedback strategy according to the device type of the target device.
[0139] Step 812: Generate fault warning information based on the result feedback strategy and send the fault warning information to the target device.
[0140] In this application, network signal sequences, network fault codes, and network topology information are combined to comprehensively analyze the network fault status of the target vehicle, thereby achieving high-precision real-time fault diagnosis in complex network environments with multiple gateways.
[0141] The following describes an embodiment of the apparatus described in this application, which can be used to execute the fault diagnosis method described in the above embodiments of this application. For details not disclosed in the apparatus embodiments of this application, please refer to the embodiments of the fault diagnosis method described above.
[0142] See Figure 9 The diagram shows a block diagram of a fault diagnosis device 900 according to an embodiment of this application. The fault diagnosis device 900 is configured in the cloud corresponding to the gateway system of the target vehicle. The gateway system includes a central brain and at least two regional gateways, specifically including:
[0143] Information acquisition module 901 is used to acquire network signal sequences and network fault codes uploaded by the central brain;
[0144] The first feature extraction module 902 is used to extract features from the network signal sequence to obtain signal temporal features; wherein, the signal temporal features are used to characterize the signal change trend of the network signal sequence;
[0145] The second feature extraction module 903 is used to extract features from network fault codes to obtain fault semantic features; wherein, the fault semantic features are used to characterize the degree of correlation between fault objects corresponding to network fault codes.
[0146] The path generation module 904 is used to generate a signal propagation path based on the network topology information of the gateway system.
[0147] The fault diagnosis module 905 is used to perform network fault diagnosis on the target vehicle based on signal timing characteristics, fault semantic characteristics, and signal propagation path, and obtain the diagnosis results.
[0148] In an exemplary embodiment, the first feature extraction module 902 described above includes:
[0149] A sequence partitioning unit is used to divide a network signal sequence into at least two signal subsequences of a preset time length;
[0150] The temporal feature extraction unit is used to input at least two signal subsequences into the signal analysis model to extract features from the at least two signal subsequences and obtain the signal temporal features output by the signal analysis model; wherein, the signal analysis model is built based on LSTM.
[0151] In an exemplary embodiment, the second feature extraction module 903 described above includes:
[0152] The text acquisition unit is used to acquire the descriptive text data corresponding to the network fault codes; wherein, the descriptive text data is used to define the fault objects corresponding to different network fault codes.
[0153] The semantic feature extraction unit is used to input network fault codes and descriptive text data into the fault analysis model, so as to extract features from the network fault codes based on the descriptive text data and obtain the fault semantic features output by the fault analysis model; wherein, the fault analysis model is built based on BERT.
[0154] In an exemplary embodiment, the fault diagnosis device 900 further includes:
[0155] The knowledge graph construction module is used to construct a knowledge graph of the gateway system with the central brain and regional gateways as nodes and the first and second communication links as edges; wherein, the first communication link is the communication link between the central brain and each regional gateway, and the second communication link is the communication link between each regional gateway.
[0156] The topology information extraction module is used to extract the network topology information corresponding to the different vehicle functions involved in the gateway system from the knowledge graph of the gateway system.
[0157] In an exemplary embodiment, the path generation module 904 is specifically used to input the network topology information corresponding to each vehicle function involved in the gateway system into the graph structure recognition model to obtain the signal propagation path under the vehicle function output by the graph structure recognition model; wherein, the graph structure recognition model is built based on GCN, and the signal propagation path includes the regional gateway associated with the vehicle function.
[0158] In an exemplary embodiment, the fault diagnosis module 905 includes:
[0159] The path feature extraction unit is used to convert the signal propagation path into a text sequence and extract features from the text sequence to obtain the signal path features.
[0160] The fault diagnosis unit is used to normalize the signal timing features, fault semantic features, and signal path features, and then perform weighted fusion processing on the normalized signal timing features, fault semantic features, and signal path features to obtain the fault probability information corresponding to the central brain and each regional gateway.
[0161] In an exemplary embodiment, the fault diagnosis device 900 further includes:
[0162] The strategy selection module is used to respond to the diagnostic result request for the target vehicle sent by the target device and determine the corresponding result feedback strategy according to the device type of the target device.
[0163] The result feedback module is used to generate fault warning information based on the result feedback strategy and send the fault warning information to the target device.
[0164] Based on the same inventive concept, embodiments of this application provide a computer-readable storage medium storing computer program instructions. When the computer program instructions are loaded and executed by a processor, they implement the steps of the fault diagnosis method described above.
[0165] Based on the same inventive concept, this application provides an electronic device, see [link to relevant documentation]. Figure 10 The diagram shows a schematic of the structure of an electronic device in an embodiment of this application. The electronic device includes one or more memories 1004, one or more processors 1002, and at least one computer program stored in the memory 1004 and executable on the processor 1002. When the processor 1002 executes the computer program, it implements the steps of the fault diagnosis method described above.
[0166] The bus architecture (represented by bus 1000) includes any number of interconnected buses and bridges, linking various circuits including one or more processors represented by processor 1002 and memory represented by memory 1004. Bus 1000 can also link various other circuits such as peripherals, voltage regulators, and power management circuits, which are well known in the art and therefore will not be described further herein. Bus interface 1005 provides an interface between bus 1000 and receiver 1001 and transmitter 1003. Receiver 1001 and transmitter 1003 can be the same element, a transceiver, providing a unit for communicating with various other devices over a transmission medium. Processor 1002 is responsible for managing bus 1000 and general processing, while memory 1004 can be used to store data used by processor 1002 during operation.
[0167] The functions described herein may be implemented in hardware, software executed by a processor, firmware, or any combination thereof. If implemented in software executed by a processor, the functions may be stored as one or more instructions or codes on or transmitted via a computer-readable medium. Other examples and embodiments are within the scope and spirit of this application and the appended claims. For example, due to the nature of software, the functions described above may be implemented using software executed by a processor, hardware, firmware, hardwired, or any combination thereof. Furthermore, the functional units may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit.
[0168] Based on the same inventive concept, this application provides a computer program product, including a computer program, which, when executed by a processor, implements the steps of the fault diagnosis method described above.
[0169] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For instance, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling, direct coupling, or communication connection may be through some interfaces; the indirect coupling or communication connection between units or modules may be electrical or other forms.
[0170] The units described as separate components may or may not be physically separate. Similarly, the components of the control device may or may not be physical units; they may be located in one place or distributed across multiple units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.
[0171] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing computer program instructions, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.
[0172] The above description is merely an embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.
Claims
1. A fault diagnosis method, characterized in that, The cloud-based gateway system applied to the target vehicle includes a central brain and at least two regional gateways. The method includes: Obtain the network signal sequence and network fault code uploaded by the central brain; Feature extraction is performed on the network signal sequence to obtain signal temporal features; wherein, the signal temporal features are used to characterize the signal change trend of the network signal sequence; Feature extraction is performed on the network fault codes to obtain fault semantic features; wherein, the fault semantic features are used to characterize the degree of correlation between the fault objects corresponding to the network fault codes; Generate a signal propagation path based on the network topology information of the gateway system; Based on the signal timing characteristics, the fault semantic characteristics, and the signal propagation path, network fault diagnosis is performed on the target vehicle to obtain the diagnosis result.
2. The method according to claim 1, characterized in that, The step of extracting features from the network signal sequence to obtain signal temporal features includes: The network signal sequence is divided into at least two signal subsequences of a preset time length; The at least two signal subsequences are input into the signal analysis model to extract features from the at least two signal subsequences, thereby obtaining the signal temporal features output by the signal analysis model; wherein the signal analysis model is constructed based on the Long Short-Term Memory (LSTM) network.
3. The method according to claim 1, characterized in that, The step of extracting features from the network fault codes to obtain fault semantic features includes: Obtain the descriptive text data corresponding to the network fault code; wherein, the descriptive text data is used to define the fault object corresponding to different network fault codes; The network fault code and the descriptive text data are input into the fault analysis model to extract features from the network fault code based on the descriptive text data, thereby obtaining the fault semantic features output by the fault analysis model; wherein, the fault analysis model is constructed based on the bidirectional encoder BERT.
4. The method according to claim 1, characterized in that, Before generating the signal propagation path based on the network topology information of the gateway system, the method further includes: Using the central brain and each regional gateway as nodes, and the first communication link and the second communication link as edges, a knowledge graph of the gateway system is constructed; wherein, the first communication link is the communication link between the central brain and each regional gateway, and the second communication link is the communication link between each regional gateway. From the knowledge graph of the gateway system, the network topology information corresponding to the different vehicle functions involved in the gateway system is extracted.
5. The method according to claim 4, characterized in that, The step of generating a signal propagation path based on the network topology information of the gateway system includes: For each vehicle function involved in the gateway system, the network topology information is input into a graph structure recognition model to obtain the signal propagation path under the vehicle function output by the graph structure recognition model; wherein, the graph structure recognition model is constructed based on a graph convolutional network (GCN), and the signal propagation path includes the regional gateway associated with the vehicle function.
6. The method according to claim 1, characterized in that, The step of performing network fault diagnosis on the target vehicle based on the signal timing characteristics, the fault semantic characteristics, and the signal propagation path to obtain a diagnosis result includes: The signal propagation path is converted into a text sequence, and features are extracted from the text sequence to obtain signal path features; The signal timing features, fault semantic features, and signal path features are normalized, and the normalized signal timing features, fault semantic features, and signal path features are then weighted and fused to obtain the fault probability information corresponding to the central brain and each regional gateway.
7. The method according to claim 6, characterized in that, The method further includes: In response to a diagnostic result request for the target vehicle sent by the target device, a corresponding result feedback strategy is determined based on the device type of the target device; Based on the result feedback strategy, a fault warning message is generated and sent to the target device.
8. A fault diagnosis device, characterized in that, The cloud-based gateway system configured for the target vehicle includes a central brain and at least two regional gateways. The device includes: The information acquisition module is used to acquire network signal sequences and network fault codes uploaded by the central brain; The first feature extraction module is used to extract features from the network signal sequence to obtain signal temporal features; wherein, the signal temporal features are used to characterize the signal change trend of the network signal sequence. The second feature extraction module is used to extract features from the network fault codes to obtain fault semantic features; wherein, the fault semantic features are used to characterize the degree of correlation between the fault objects corresponding to the network fault codes; The path generation module is used to generate a signal propagation path based on the network topology information of the gateway system; The fault diagnosis module is used to perform network fault diagnosis on the target vehicle based on the signal timing characteristics, the fault semantic characteristics, and the signal propagation path, and obtain the diagnosis results.
9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer program instructions that are loaded and executed by a processor to perform the operations described in any one of claims 1 to 7.
10. An electronic device comprising a processor and a memory, characterized in that, The memory stores computer program instructions that can be executed by the processor, and when the processor executes the computer program instructions, it implements the instructions of the method as described in any one of claims 1 to 7.